Fractal Analysis of CT Perfusion Images for Evaluation of Antiangiogenic Treatment and Survival in Hepatocellular Carcinoma

Abstract

Tumor vascular heterogeneity is a recognized biomarker for cancer progression. Our purpose was to assess the tumor perfusion heterogeneity during antiangiogenic therapy in hepatocellular carcinoma (HCC) by means of fractal analysis on computed tomography perfusion (CTP) images. Twenty-two patients (15 men and 7 women; mean age: 61.5 years) with advanced HCC underwent CTP at baseline and 2 weeks after administration of bevacizumab. Perfusion maps of blood flow (BF) were generated by the adiabatic approximation to the tissue homogeneity model with a motion registration, and fractal analyses were applied to gray-scale perfusion maps using a plugin tool on ImageJ software (NIH, Bethesda, MD). A differential box-counting method was applied, and the fractal dimension (FD) was calculated as a heterogeneity parameter. Patients were grouped into favorable progression-free survival (PFS) group (PFS>6 months, 11 patients) and unfavorable PFS group (PFS≤6, 11 patients). After 2 weeks of antiangiogenic therapy, the BF decreased significantly (P < .0001), whereas the FD showed no significant change (P = .69). The percent change of the FD in tumor BF was significantly different between patients with favorable PFS and those without (-2.52% vs. 3.72%, P = .01), whereas the change of tumor BF showed no significant difference between them (-28.93% vs. -25.47%, P = .64). In Kaplan-Meier analysis, patients with greater reduction in the percent change of FD and lower baseline FD in tumor BF showed significantly longer overall survival (P = .009, P = .005). Fractal analysis of tumor BF can be a biomarker for antiangiogenic therapy.